In an increasingly connected world, the internet of Things (IoT), Cloud and Fog computing are a major asset allowing to overcome previously inconceivable limits in terms of innovation. However, Fault Tolerance remains a major challenge for assuring IoT systems dependability. In order to tackle this issue, we propose a generic microservice architecture called FaTMA (Fault Tolerance- Microservice Architecture for IoT) permitting detection of Things failures by providing continuous and real-time monitoring of their states. In addition, it offers mechanisms to strengthen the reliability of the designed systems. We adopt Bigraphical Reactive Systems (BRS) as formalism to define a formal model that describes architectural elements of different IoT system layers and their behavior. It provides a clear separation between the various microservices controlling this system type and their side effects. Indeed, the execution of the proposed model, through BigraphER tool, permits to simulate and analyze different failure scenarios as well as their restitution strategies.